<RaYmAn>
yeah, but if you take 0xf0000000 and add to 0x40000100, you get 0x130008000
<RaYmAn>
which is outside the valid range =P
<RaYmAn>
also, that's the base of the DRAM registers :/
<RaYmAn>
hmm
<RaYmAn>
ok, so sdram starts at 0xc0000000
<RaYmAn>
> virt2phys 0xc0000100
<RaYmAn>
Physical address 0x40000100
<mnemoc>
good point
<RaYmAn>
I guess this is just one of those things you have to accept as truth
<RaYmAn>
lol
<mnemoc>
:)
RITRedbeard has quit [Read error: No route to host]
RITRedbeard has joined #arm-netbook
<RaYmAn>
mnemoc: at some point, would you be able to maybe enable kexec + atags (doesn't matter whether kexec works, since I just need the atags stuff) on some a10 device and make a dump for me? :S It seems that without that feature enabled, kernel overwrites the atags
<Boulet>
i cleaned up my test1 a little bit, and now it returns to fel
<Boulet>
but i also removed uart init from it
<RaYmAn>
Boulet: have a look at hno's cleaned up variant
<RaYmAn>
that's missing the return to fel
<Boulet>
all the uart and printf are really not necessary, it was just during dev
<RaYmAn>
no, but it's nice to have still
<RaYmAn>
so for return to fel, did you just get rid of entry.S and put main at 0x0?
<mnemoc>
RaYmAn: on A10 I can give you the ATAGs given by hno's uboot. on A13 I can't yet because I don't have userspace yet
<RaYmAn>
mnemoc: a10 is fine - I just need the binary version
<RaYmAn>
I guess I could theoretically figure out how u-boot generates them and recreate that :S
<mnemoc>
give me 15m
<mnemoc>
CONFIG_ATAGS_PROC=y <--- that one?
<RaYmAn>
yup
<RaYmAn>
it requires the kexec config to work
hipboi has quit [Quit: 暂离]
<mnemoc>
my uart dongle decided to go on strike again :<
<WarheadsSE>
I'll have to try an updated kernel on my mele.. X borked :/
Almamuerta has joined #arm-netbook
Boulet has quit [Ping timeout: 245 seconds]
Boulet has joined #arm-netbook
Boulet has quit [Ping timeout: 244 seconds]
kaspter has quit [Ping timeout: 276 seconds]
kaspter has joined #arm-netbook
<mnemoc>
RaYmAn: my mele's uart is acting seriously weird... I have a loop echoing "test\n" to ttyS0, which according to stty is set at 115200 (as expected) but on the other end I receive "WH.WHY."...
<mnemoc>
also set at 115200... and same with 3 dongles
<mnemoc>
RaYmAn: I'll try updating the uboot of the SD to support uEnv.txt and try to start the test kernel that way
<RaYmAn>
odd
<mnemoc>
WH.WHYY.WH.WHYYY.WH.WH.WH.
<mnemoc>
looks kind of funny
Boulet has joined #arm-netbook
<RaYmAn>
mnemoc: test1 returns to fel now. (current git). fel write only writes first 8kb of what you give it, but it writes those correctly to ram ;)
<RaYmAn>
mnemoc: I verified by dumping with jtag, everything after 8kb is 'random' rather than the data (exactly 8kb mark)
<RaYmAn>
So fel tool probably does something wrong
<RaYmAn>
I might look into it later
<RaYmAn>
but once that is fixed, we should be able to boot kernel from fel
<mnemoc>
RaYmAn: if the nand driver is y, yes, /dev/nand* are detected immediately
<mnemoc>
uh, hipboi switched to freeshipping for the uhost thingie
Quarx|2 has quit []
Quarx|2 has joined #arm-netbook
Quarx|2 is now known as Quarx
Quarx has quit [Client Quit]
Quarx has joined #arm-netbook
<RaYmAn>
mnemoc: hrm. I'm getting [ 0.950000] [ccmu] set nfc clock rate to 60000000 failed!
<RaYmAn>
:(
<Turl>
nfc hmm
<RaYmAn>
it's nand stuff
<Turl>
RaYmAn: :/
<RaYmAn>
mnemoc: any chance you have time to try out fel kernel boot on that a13 olinuxino? :>
<Turl>
RaYmAn: while I was working on cpufreq I've seen some similar messages
<Turl>
when I failed and wrote invalid stuff :P
<RaYmAn>
I guess maybe it's expecting some clocks to have been set already by bootloader
<Turl>
probably
<mnemoc>
RaYmAn: ah, 1m
<Turl>
RaYmAn: a10 or a13 btw?
<RaYmAn>
Turl: 13
<Turl>
might be that then
<mnemoc>
RaYmAn: we "fixed" that in sun4i's
<mnemoc>
in 3.4 both drivers are merged but not in 3.0
<RaYmAn>
ah
<mnemoc>
vi
<RaYmAn>
nano
<RaYmAn>
:P
<mnemoc>
in
<mnemoc>
er
<mnemoc>
I can'\t type
<mnemoc>
f* f* f*
* mnemoc
breaths
<Turl>
nano ftw ;)
<mnemoc>
in drivers/block/sun5i_nand/nfd/nand_blk.c
* RaYmAn
consistently uses nano
<mnemoc>
commented #define USE_SYS_CLK out
<mnemoc>
comment #define USE_SYS_CLK out
<Turl>
RaYmAn: I do too
<RaYmAn>
my colleagues laugh at me :(
<mnemoc>
i would too
<RaYmAn>
mnemoc: that's it? :)
<Turl>
yeah I have a friend who laughs because I don't use vi :P
<mnemoc>
RaYmAn: yes, just comment that #define out
<RaYmAn>
I switch systems too much to use vi(m), emacs or similar.
Almamuerta has joined #arm-netbook
<mnemoc>
vi is pretty consistent across systems
Almamuerta has quit [Client Quit]
<mnemoc>
emacs is certainly different everywhere
<RaYmAn>
yeah, but it's not really that good on mac or windows
<Turl>
I'd s/consistent/consistently weird/ :P
<RaYmAn>
I use Visual Studio in part of my work as well
<RaYmAn>
Having to remember all the special keys for VI or emacs is just too much trouble for me =P
* mnemoc
won't be able to look at RaYmAn in the same way again
<RaYmAn>
lol
<RaYmAn>
I like doing different things :)
<mnemoc>
*cough*
<Turl>
I just hope you don't program VB6
<RaYmAn>
If I did, I'd certainly never have taken this job
<RaYmAn>
.Net 4.0 and C#
<RaYmAn>
that's half or so
<RaYmAn>
the rest is python, wsgi and postgresql
<RaYmAn>
mostly
<RaYmAn>
a bit of C too
Quarx has quit []
<RaYmAn>
mnemoc: that at least made partitions appear :)
<RaYmAn>
PHY_PageReadSpare : too much ecc err,bank 0 block 0,page 1
<RaYmAn>
:S
<RaYmAn>
but I guess that's not really an issue
<RaYmAn>
is it common to end up with uhm, "vibrating" coloured bars when this thing fails to boot? :P
<mnemoc>
no idea, never booted a device with display :p
<RaYmAn>
lol
<RaYmAn>
olinuxino uses lcd pins though :P
<RaYmAn>
I get this when booting that kernel (from memory) skumler.net/colorbars.jpg
<RaYmAn>
at least that means it initializes the LCD correctly :P
<Turl>
RaYmAn: yeah I also get the ecc thing on my A10
<RaYmAn>
'k
<Turl>
but no issues so far other than it being printed
<WarheadsSE>
trying 3.0.39 for ALARM
<RaYmAn>
I think me or someone else is going to have to hack together an atag creator
<RaYmAn>
if we want to be able to properly boot over fel
<RaYmAn>
mnemoc: Ijust noticed I'm using an old fork on my github, lol (kernel)
<RaYmAn>
might help to update :>
<mnemoc>
:)
<RaYmAn>
I just realized that if I can get a minimal ramdisk linux (with console over uart) going from this, I can dump nand completely without ever having flashed anything to device
<WarheadsSE>
heh
<WarheadsSE>
mnemoc: remember when I was having those issues with xorg
* RaYmAn
looks around for volounteers for writing a createatags program
<WarheadsSE>
Also, funny as hell .. => configure exorg to use fbdev & /dev/fb0 ;)
<WarheadsSE>
load up the mali drivers, then start X
<WarheadsSE>
"interesting" behavior
<Turl>
isn't that how everyone is doing it WarheadsSE?
<WarheadsSE>
i don't think so
<WarheadsSE>
I compiled w/ the mali driver
<Turl>
RaYmAn: I might be able to help :)
<RaYmAn>
ooh, a volounteer! :D
<Turl>
RaYmAn: do you have docs for the atag format?
<RaYmAn>
Turl: no, but I have a program that reads a binary atag file and dumps the contents
<RaYmAn>
it has a statically compiled version of the app too - and included an atag dump from mele a1000
<Turl>
RaYmAn: so what's the expected behaviour?
<Turl>
(of the createatags program)
<RaYmAn>
Turl: in particular, it needs to drop in the ATAG_CORE as first tag, then an ATAG_mem based on user params, optionally an ATAG_CMDLINE with cmdline (usually max 1024 chars) and an INITRD2 ATAG based on uhm, providing offset and a ramdisk file (or just manually specify size)
<RaYmAn>
the rest of the atag's aren't really required for this purpose
<RaYmAn>
tbh, it doesn't need to be fancy - I'm planning to write wrapper scripts around it anyways. (e.g. fel-boot.sh --kernel somekernel --ramdisk someramdisk --board a13 ) and let that handle uploading the right stuff to memory and passing suitable parameters
<mnemoc>
hno: once it learns to initialize dram ;-)
<mnemoc>
well... i mean generically
<hno>
what I am talking about above is merely 3 instructions to load r0, r1, r2 with proper values so kernel do not need to override at startup.
<hno>
SPL will soon be as generic as boot0 I hope. I.e. needing a matching script.bin and running a tool to copy the needed parameters into the bootloader.
<mnemoc>
i was wondering how to reduce code duplication between this mini-spl and spl
<mnemoc>
mini-spl doing dram, minimal atag, and r0,r1,r2 .... then the spl part that knows to read nand/mmc on any sunxi, then fancy full uboot, then linux
<mnemoc>
or, when using fel, linux directly after mini-spl
<mnemoc>
that would allow a larger spl i believe
<mnemoc>
because dram is already initialized
<mnemoc>
does uboot allow something like that?
<mnemoc>
back in ~1h
tuliom_ has joined #arm-netbook
tuliom_ is now known as tuliom
<hno>
I am not very fond of the idea of SPL trying to guess the DRAM parameters.
<hno>
The whole clock & dram setup code can be shared by fel and spl booting.
<xxiao>
someone used pogoplug's ulimit storage space, $19.99 per year for ulimited space? that beats everyone else, too good to be true
<WarheadsSE>
xxiao: wut
<WarheadsSE>
Ah, the family
<Turl>
mnemoc: :D
<WarheadsSE>
basically mini-team
<WarheadsSE>
the "unlimited" is your device.
<Turl>
mnemoc: we got gold :D
<xxiao>
i hook 4 usb 1TB on it
<WarheadsSE>
that'd work
<xxiao>
it's idle on my desk for a long while but now i want to use it, then found out pogoplug's storage is too good to be true
<WarheadsSE>
its not cloud storage!
<xxiao>
WarheadsSE: it is, they sync your local disks to their cloud storage, $19.99 per year
<WarheadsSE>
"Continuous backup of your family’s computers and mobile devices even when you’re away from home" means the software will backup your phone -TO YOUR DEVICE-
<xxiao>
i think you're right, shoot
<WarheadsSE>
nah, shit, the dude hat exploits the crap out of their devices and services is right ...
<xxiao>
i got distracted by this before i decide to load debian/archlinux/openwrt to hack it
<WarheadsSE>
The only bit of this is the *multi-account* part
<xxiao>
ic
<xxiao>
debian-on-it(or openwrt, if i can put that to NAND instead of wasting a usb port for debian image)
<xxiao>
mount S3 as FUSE to do 'cloud-backup' along with local storage
<WarheadsSE>
get real performance, put sata on it ;)
<xxiao>
should be very doable
<xxiao>
it's mostly for archiving and backup, speed is not that unbearable
* xxiao
is still waiting for self-powered-eSATA pogoplug-alike cheap boxes
<WarheadsSE>
You're expecting a device like this to have eSATAp
<xxiao>
not sure how much esatap disks are
gimli has quit [Quit: Verlassend]
<arokux_h>
I've successfully booted 3.0.39+, not a big deal for you guys, but I'm happy, since I've found a bug in scripts/depmod.sh if kmod was compiled with --with-rootprefix=<some_directory>
pwhalen has quit [Read error: Connection reset by peer]
<Turl>
RaYmAn: ping
<Turl>
I've got the atag generator thing working :)
pwhalen has joined #arm-netbook
mnemoc_ has joined #arm-netbook
arokux_h has quit [Remote host closed the connection]
mnemoc_ has quit [Client Quit]
mnemoc_ has joined #arm-netbook
<hno>
xxiao, backups which are stored in your home is fail. If your home burns down or you hae a burgler stealing your stuff then all is gone.
<hno>
Turl, great. Please make sure it finds it's way into sunxi-tools.
<Turl>
hno: yep :) need to ask RaYmAn what license does he want me to stick on his code and we'll have dumpatags + atags generator
<xxiao>
hno: right. i'm thinking about truecrypt local pogoplug, S3 remotely dual copy
<xxiao>
just built openwrt for pogoplug
* hno
have a nas kind of device at his mother in law.
<hno>
and don't care about encryption. Sensitive stuff is kept encrypted at file level.